Coordination chemistry of microbial iron transport compounds. 11

Jan 1, 1979 - Journal of the American Chemical Society. Harris, Carrano, Cooper, Sofen, Avdeef, McArdle, Raymond. 1979 101 (20), pp 6097–6104...
0 downloads 0 Views 712KB Size
Carrano, Cooper, Raymond

1 Equilibria of

599

Ferric Rhodotorulate

Coordination Chemistry of Microbial Iron Transport Compounds. 1 1. Solution Equilibria and Electrochemistry of Ferric Rhodotorulate Complexes Carl J. Carrano, Stephen R. Cooper, and Kenneth N. Raymond* Contribution from the Department of Chemistry,Unic.ersity of California, Berkeley, California 94720. Receiced July 18, 1978

Abstract: Rhodotorulic acid (HzRA) is a dihydroxamic acid which is produced in high yields by yeast as an iron transport agent (siderophore). The stability constants of the ferric complexes of RA have been determined potentiometrically and spectrophotometrically at 25 "C. The predominant species at neutral pH is the dimer, FezRA3, in which each iron is bound to three hydroxamate groups. Below pH -3 this complex dissociates into the monomer, FeRA+, in which each iron is bound to two hydroxamates. The formation constants for Fe2RA3 and FeRA+ from Fe3+ and RA2- are log p = 62.2 (1) and 21.99 (6), respectively. The values of the acid dissociation constants are moderately correlated with pK,l = 8.49 (3) and pK,z = 9.44 (3). The electrochemical behavior of the ferric RA complexes was examined via cyclic voltammetry, and the pH dependence of the redox potentials determined. The reduction of Fe2RA3 is reversible and involves one electron per iron. That the two Fe(II1) centers (9-10 %, apart) do not affect the redox behavior is consistent with ESR evidence which shows no interaction of the two ions. At pH 7 and millimolar concentration of free ligand, ferrous, and ferric species the formal potential Go = -0.279 V vs. NHE. The numerical values of the formation constants of polyhydroxamate siderophores show little if any chelate effect. This observation and its implications are discussed.

Introduction Low molecular weight microbial iron sequestering agents known as siderophores have been found to be among the strongest chelators of ferric ion known2 These natural products have evolved as transport agents and scavengers for iron, an essential element for growth, in a host of bacteria, fungi, and mold^.^-^ Siderophores fall generally into two groups, either hydroxamates or catecholates, both of which form exceedingly stable octahedral, high-spin complexes of iron(II1). Because of their great ability to sequester and hold iron (many are thermodynamically able to remove iron from the mammalian iron transport protein transferrin), some of the siderophores have found use in the treatment of acute iron poisoning and chronic iron poisoning such a s occurs in the transfusion therapy of Cooley's The most used clinical agent in such poisoning cases is Desferal (deferriferrioxamine B). This agent binds iron with an overall stability constant of 30.6,* and has recently been licensed for medical use in the U.S.A. Although other factors come into play, it is clear that the stability constant is important in determining which chelates will be useful drugs. The stability constants for several of the siderophores are known, primarily the ferrioxamines and ferrichromes.2 One of the general types about which little has been known is rhodotorulic acid, isolated from Rhodotorula pilimanae and related yeasts (Figure This dihydroxamate diketopiperazine has already been found to be more effective than Desferal as an iron chelating drug and has entered limited clinical trials,1° although knowledge of even the composition of its iron complex has not been obtained until recently.' Owing to its potential usefulness as a drug and in the course of our investigation into the coordination chemistry of this novel siderophore, we have examined the equilibrium chemistry of rhodotorulic acid via potentiometric titration, spectrophotometric titration, and electrochemistry. The results of these investigations are reported here. Experimental Section Ferric stock solutions (approximately 0.1 M in 0.1 M HC1) were prepared from ferric chloride hexahydrate or ferric nitrate (0.1 M in 0.1 M "03) and standardized both spectrophotometrically and by ignition to the oxide (900 O C , 12 h). Rhodotorulic acid was prepared as described by Neilands et al.9

and recrystallized at least six times before use. The product had IR, NMR, and melting point identical with the reported values. The purity of the ligand was also established by titration with standard KOH. Potentiometric Titrations. The potentiometric titrations were performed as described by Raymond et al." Spectroscopic Measurements. Visible absorption spectra were measured using a Cary Model 118 UV-vis spectrophotometer. Solutions were prepared by mixing ligand and metal at molar ratios of 5:l in individual portions. Small amounts of acid or base titrant were added and the pH and spectra were recorded. The system was kept at 1.0 M ionic strength via addition of KNO3. In order to determine the equilibrium constant for the reaction Fe3+

+ H2L

FeL+

+ 2H+

it h a s necessary to adjust the acidity to p H values near zero decrease M) to 1. At such low values of pH the L / M ratio ( [ F e l ~= 4 X the glass electrode response is erratic and it was necessary to resort to calculations of [H+] based on total acid and base in the system. In such cases the reactions were carried out in 0.1 or 1 M HCI and the [H+] was calculated after additions of standard base. Computational Techniques. A computer program, STBLTY, was used for computation and refinement of acid dissociation and stability constants as previously described." Electrochemistry. Cyclic voltammograms were recorded using three-electrode circuitry with a hanging mercury drop working electrode, platinum wire auxiliary, and saturated calomel reference electrode (SCE) and are uncorrected for liquid junction potentials. All solutions for electrochemical examination were 1.O M in KC1 and 0.05 M each in potassium phosphate and sodium borate. Triangular waves were generated by the Princeton Applied Research (PAR) 175 programmer in conjunction with the PAR 380 coulometry system, which includes the PAR 173 potentiostat. Controlled potential elec. trolyses were conducted on a stirred mercury pool. Iron electrodes for potentiometric measurements were prepared by electroplating iron onto a copper wire from a stirred solution of ferrous chloride (28 g/100 mL of 0.01 M HCI) at 70 OC, using an analytically pure iron wire electrode. Potentiometric measurements were made with a Brinkmann pH-102 pH meter in the mV mode.

Results Ligand Titration. The pK, of the hydroxamate portion of free rhodotorulic acid was originally reported by Neilands et al.,9who found a value of -9.3 for the pK, but reported that the titration curve was skewed and mathematically incompatible with a single pK,. In our examination of the ligand-only titration data, both hydroxamate groups and the deprotonation 0 I979 American Chemical Society

Journal of the American Chemical Society / 101:3

600

/ January 31, 1979

U

H Figure 1. Structure of rhodotorulic acid. PH

4

3

h

2 Figure 3. Metal-ligand titration curves for ferric rhodotorulate: (-), 1 .O mM [Fe3+],L/M = 10.0; (- -), 3.7 mM [Fe3+],L/M = 2.6; (- - -), 1.9 mM [Fe3+],L/M = 10.0; (. . . .), 1.9 mM [Fe3+],L/M = 5.2.

I 1

,

'

Fe (IJI) -Rhodotorulate

C

425 nm

A 8

IO

I

I

12

\

I

PH Figure 2. Ligand deprotonation curve of free rhodotorulic acid, with representing average number of protons bound per ligand.

of the ring amide nitrogens at high pH were considered. As expected, the hydroxamate protons titrated in the region between pH 8 and 10. The data could be fitted equally well by assuming that the two hydroxamate protons had similar but slightly different pK,s or that they both had a single pK,, the average of the two above. We chose the former case and obtained values of 8.49 (3) and 9.44 (3). By examination of Bjerrum12 plots the deprotonation of the ring amide nitrogens was seen as the pH was raised above 10 (Figure 2 ) . The values calculated for these constants were 12.7 (1) and 13.2 (est); the latter value, an estimate, was not refined. The overall fit of the experimental data to the refined constants is excellent-the goodness of fit is 0.72.13 Ferric Rhodotorulate. Potentiometric titration data at several ligand to metal ratios and total metal concentrations revealed a sharp break at 3 equiv of base per mol of iron (Figure 3). This indicates that each iron is being coordinated by three hydroxamate groups with the concomitant release of three protons. The formation curves were surprisingly simple with Ti (the average number of ligands bound per Fe) 1.5 over a wide range of pH values.I1 This confirms our original formulation of the complex as Fe2RA3 throughout most of the pH range.' The stepwise stability constants of interest, K1 and K2, are defined as follows:

-

Fe3+ +13A22FeRA+

KI

FeRA+

K2 + RA2- + Fe2RA3

(1)

(2)

Unfortunately it proved to be impossible to refine or deduce log K1 from the potentiometric data, since fi never became less

350 nm

I

2

3

4

5

6

7 PH

9

9

IO

I1

12

Figure 4. Plots of extinction coefficient of ferric rhodotorulate vs. pH at various wavelengths.

than one in the pH range accessible by this technique. In fact the simple 1:1 complex detected spectrophotometricallyproved to be stable down to a pH of 0 and below. Alternative ways to determine initial values of this constant were sought for use in calculation of the titration results. It proved possible to determine these spectrophotometrically. Spectrophotometric Titration. A spectrophotometric titration revealed changes in the visible absorption spectra with pH characteristic of only two species being present in solution. As the pH was lowered from 7.0 to 2.0 the A, shifted from 425 to 480 nm with an isosbestic point observed at 488 nm. Plotting the extinction coefficient vs. pH (Figure 4), as described by McBryde,14revealed two well-separated plateaus between pH 2.0 and 11.0 indicating that K1 and K2 differ sufficiently to allow their determination. Another plateau was found between pH 0 and 2 which allowed us to determine K I . For the monomer-dimer equilibrium, eq 2, the data were analyzed for two wavelengths, 550 and 425 nm, at several pH values each. For the free metal-monomer equilibrium, eq 1, only a single wavelength, 480 nm, was analyzed at seven pH values. The extinction coefficients for the pure monomer and dimer were found from the extinction coefficient vs. pH plots. Above pH 4.0 all the complex exists as the dimer (E425 2700 L cm-l

Carrano, Cooper, Raymond

60 1

/ Equilibria of Ferric Rhodotorulate -700r

I

Fe-RA

0.6 - E v s SCE

0.4

0.8

mol-'). Between pH 1.O and 2.0 all the iron is present as the monomer (6480 1750 L cm-l mol-]). It was assumed that the extinction coefficient for free iron was zero in the wavelengths of interest. It was also necessary to correct for the equilibrium Ki + C1- + FeCI2+

Since iron binding by chloride in 1 M HC1 becomes significant, log Ki' = 0.63.15 The results for K1 were obtained a second time a t 0.1 M ionic strength in 0.1 M HC104 by reducing the concentration of both metal and ligand. The value of log K1 a t 1 M ionic strength was 21.55 (8) and a t 0.1 M ionic strengtth 21.99 (6). The value of log K2 a t 1 M ionic strength was determined to be 19.18 (3). The overall stability constants PI and p3/2 are defined as follows:

+FeRA+ 2Fe3+ + 3RA2- +Fe2RA3 +

Fe3+ RA2-

7

8

I

9

I

('0

Figure 6. The pH dependence of Fe211'RA3 reduction potential by cyclic voltammetry on a hanging mercury drop electrode. [Fe(llI)] = 1 mM, [RA] = 20 mM in 1 M KCI, buffered with 0.05 M potassium phosphate, 0.05 M sodium borate. The solid line is the reduction potential calculated from the equilibrium constants reported here.

Figure 5. Cyclic voltammogram of Fe2"'RA3 on a hanging mercury drop electrode. [Fe(lll)] = I m M . [RA] = 20 m M in 1 M KCI with 0.05 M potassium phosphate, 0.05 M sodium borate buffer, pH 7.0. Scan rate 100 mV/s.

Fe3+

I

PH

pH 7.0

0.2

I

Bl

(3)

@3/2

(4)

Thus PI = K1 and p3/2 = K12K2. This gives log PI = 21.55 (8) and log p3/2 = 62.28 (12). The use of (at p = 0.1 M) as a fixed parameter made possible refinement of p3/2, using the STBLTY program, to yield a value of 62.2 (1). Using these two constants the overall fit of the potentiometric titration data with the calculated values was excellent, with a final goodness of fit of 1.34, which confirms our formulation of the FeRA F= Fe2RA3 equilibrium. Electrochemistry. Cyclic voltammetric (CV) examination of Fe2RA3 solutions on a hanging mercury drop electrode yielded reversible or quasi-reversible waves over the pH range 5.5-10.5 for an RA:Fe(III) ratio of 20:l (Figure 5). The half-wave potential in this system was observed to become more negative with increasing pH until a limiting value of -656 mV vs. SCE was approached asymptotically (Figure 6). At lower ligand to metal ratios a sharp spike a t -0.33 V was observed on anodic scans following the reduction wave. The spike, which had the sharpness characteristic of an adsorption process, decreased in current with increasing R A concentration, with concomitant increase in the reversible anodic wave. At 20: 1 RA:Fe ratio no trace of the adsorption wave was present over the entire p H range. Preparative reduction of Fe2RA3 on a stirred mercury pool electrode yielded a colorless solution of the ferrous complex, which could be reversibly reoxidized to the original orange

solution of the ferric complex, and in which an iron wire exhibited a formal potential (corrected to unit concentrations of R A dianion and Fe2RA3) of -663 ( I O ) mV vs. the normal hydrogen electrode ("E).

Discussion Electrochemistry. The reduction waves observed for FezRA3 are consistent with simple reversible one-electron reduction of the ferric ion to the ferrous state. The preparative electrochemical reduction of Fe2RA3 to colorless solutions (lacking the characteristic ligand to metal charge transfer band of the ferric complex) demonstrates that simple reduction of the ferric ion is occurring. At 20:l RA:Fe ratio the cyclic voltammogram exhibits a peak separation of 60 mV a t p H 8.3, increasing to approximately 75 mV a t the limits of the p H range 5.7-10.5. For an electrochemically reversible couple a t 25 "C the predicted separation is 5 9 / n mV, where n is the number of electrons transferred.I6 Since the separation was not observed to be scan rate dependent, we attribute the observed behavior to that of an electrochemically reversible one-electron couple. For such couples the observed half-wave potential may be identified with the thermodynamic reduction potential, to within an excellent approximation. It has been claimed previously that the cyclic voltammogram of a dimer (nl = n2 = 1) should exhibit a 42-mV peak separation when the two electroactive centers have the same redox p0tentia1.I~However, Anson and coworkers have recently shown that the analysis leading to the above prediction neglects statistical factors associated with the indistinguishability of electroactive sites. Inclusion of these considerations indicates that a peak separation of 60 mV is expected, just as in the single site case, but with twofold greater current.18 Our CV results are in complete accord with the predictions of Anson and co-workers. Evidently the Fe(II1) ions are sufficiently se arated (molecular models show that the separation is 9-10 ) that they do not interact and are reduced independently. In support of this explanation, our EPR studies revealed no difference between the spectrum of Fe2RA3 and FeAlRA3 (produced by addition of a very small quantity of a ferric salt to a solution of Al2RA3). Decreasing the p H below 7 causes the observed half-wave potential to become more positive a t a rate of approximately 56 mV/pH unit. This observation is consistent with protonation of the reduction product (the ferrous complex) by one proton per electron added. Estimating the stepwise formation constants of the ferrous R A complex by those of acetohydroxamate reveals that the predominant form of the ferrous complex changes from the Fe211RA32- dimer a t high p H to the monomeric, "head to tail" FeIIRA complex a t p H -8 (Figure 7 ) . Taking account of the reaction

w

Fe211RA32- * 2FeIIRA

+ RA2-

602

Journal of the American Chemical Society

/ 101:3 / January 31, 1979

Table 11. Overall Stability Constants of Hydroxamate Siderophores siderophores ferrioxamine B (Desferal) ferrioxamine E ferrioxamine DI ferrichrome ferrichrome A ferrichrysin my cobact in a rhodotorulic acid enterobactin

PH

Table 1. pH Dependence of Ferric Rhodotoruiate Cyclic Voltammetrya (VS.

5.76 5.91 6.06 6.24 6.44 6.78 7.01 7.32 7.54 7.83 7.97 8.30 8.62 9.50 10.49

49 5 515 532 549 564 586 599 617 627 639 643 649 652 655 655

SCE)

EPbSd(VS.SCE)

AEb

517 529 545 556 571 593 60 1 618 628 635 638 640 652 656 659

73 78 70 72 67 65 68 65 65 70 65 60 67 68 73

a [Fe(III)] = 1 mM; [RA] = 20 m M ;scan rate 100 mV/s. bPeak separation, mV.

one can calculate the free ferric and ferrous ion concentrations from the distribution curves in Figure 7. By employing the Nernst equation, the pH dependence of the Fe211’RA3 reduction potential can be calculated from the ratio of free Fe(II1) to free Fe(I1). The results of this calculation are summarized in Table I, with the CV data. The limiting high pH reduction potential of -659 mV vs. SCE (-41 5 mV vs. N H E ) corresponds to the reduction of the ferric dimer to the ferrous dimer, Fe211RA32-. This potential is related to the stability constants KO, and Kred of the ferric and ferrous forms by E

,Zcomplex

59 log __ - E 1/2aquo = K O ,

2 Kred Given KO,(= p3/2 from the titration results for Fe3+),Kred (= p3/2 for Fe2+) is calculated to be From the formal potential of an iron wire immersed in a ferrous-RA solution, an independent estimate of the stability constant of the ferric dimer can be made from the Fe(III)/ Fe(I1) potentials. With suitable signs and coefficients one obtains the free energy change for

2Fe(III)

ref 1 1 1 1 19 1 20 this work 21

a Mycobactin contains two hydroxamate groups; the third coordinating group is a phenolic hydroxyl and the nitrogen of an oxazoline ring. The stability constant is an estimate from the data given by Snow. bEnterobactin is not a hydroxamate siderophore but rather a catecholate; it is included as a comparison. CEstimate;see text. d Estimate.

Figure 7. Distribution curve of the various species present at equilibrium following SO% reduction, as a function of pH ([Fe2’]~ = S X M, [ F e 3 + ]=~ S X M, [RA] T = 20 X M).

PH

log R i 30.6 32.5 30.8 29.1 32.0d 30.0 31.5d 62.3 (31.2)c 52.0

+ 3RA2-=+ Fe3II1RA3

This potential, -663 mV vs. NHE, was not highly repro-

ducible, but gives a calculated value for log KO, of -57, which is consistent with the more precise potentiometric result of 62.2. Equilibria. Using a variety of equilibrium techniques we have been able to determine the stepwise and overall stability constants for the formation of iron(II1) rhodotorulic acid complexes. As expected this hydroxamate siderophore forms extremely stable complexes with ferric ion and only very weakly binds ferrous ion (as determined electrochemically). This study has confirmed that the predominant species at neutral pH can be formulated as Fe2RA3. This dimer exists over the range from pH 3 (where the monomer becomes predominant) to pH 12 (where iron precipitates as the hydroxide), and thus it is this complex that acts as the iron transport agent in the yeast Rhodotorula pillimanae. The monomer, FeRA+, is extremely stable toward acid dissociation; it is only about 50% dissociated at pH 0 when the ligand concentration is millimolar. The magnitude of log K1, 21.55, indicates that both hydroxamate moieties in the single RA molecule are attached to the iron. For acetohydroxamate, log K1 = 11.42, log K2 = 9.68, and log K3 = 7.23.2The value we report as log K1 for RA is substantially higher than what one would expect for a single hydroxamate group attached to Fe(II1) (expected to be -10-12) but is exactly the value expected for two hydroxamate moieties attached to iron (p2 for acetohydroxamic acid is 21.10). This supports the characterization of the monomeric FeRA complex as a monocation in which both hydroxamate residues of one RA bind to iron.’ We were unable to observe a stepwise dissociation of the monomer to free iron in any experiments, although we did observe a slight pH dependence for the calculated values of K1. By way of comparison, several of the known values of the overall stability constants of the siderophores are listed in Table 11. It is impossible to compare directly the values for rhodotorulic acid with those of the other siderophores, since the equilibrium constants have different units. However, a simple comparison is possible by determining the overall constant “per iron” for which log is 112 the value of @3/2 (vide infra). Some General Comments on Hydroxamate Siderophore Equilibria. Table I11 summarizes equilibrium data for mono-, di-, and trihydroxamic acids. In Table 111 each hydroxamate functional group is represented as L, such that a dihydroxamate (e.g., rhodotorulic acid) is designated as H2LL, while a trihydroxamic acid (for example, ferrichrome) is designated H3LLL. Two interesting features follow: (1) The pK, values are approximately the same for the first, second, or third proton being removed. (2) The magnitude of the formation constant for an Fe(hydroxamate), complex is the same whether the n hydroxamate groups are separate or are part of one molecule.

Carrano, Cooper, Raymond

1 Equilibria of Ferric Rhodotorulate

603

Table 111. A Comparison of Relative Stepwise Formation Constants for Hydroxamate Ligands compd lacetohydroxamic acid (HL)

PKd

log p values FeL FeL2

FeL3

ref

9.35

11.42

21.1

28.3

2

8.49a 9.44

21.6

31.1b

this work

OH 0

I I

HN-C-CH3

(H,LL)

rhodotorulic acid

ferrioxamine B

(HjLLL)

8.39 9.03 9.10

21.8

30.6

22

ferrichrome

(H,LLL)

8.1 1 9.00 9.83

20.1

29.1

8

OH 0

I 1I

~-(-C-CH~-NH)~-~C-CH(CH~'-C-CHJbNH),-

II 0

II

0 0

Correlation coefficient, 0.66. The formation constant for '/2Fe2RA3 (= / 3 3 / ~ ' / ~ ) .

These two statements are equivalent to saying that the free energy per hydroxamate group is constant for these polyhydroxamate chelating agents and their ferric complexes at a standard state of 1 M . This results in the complete absence of any apparent "chelate effect" in comparing formation constants for the poly- or mOnohydroxamate ligands.23The following thermodynamic cycle summarizes these results for the formation of tris complexes. Similar cycles for the formation of mono or bis complexes would have the same form, with AG for interconversion of related species also 0:

-

+Fe3+

3/,H,LL

-L

1 p = o 3HL

1

A.G

3H+ + 3/,LLz-

1 1

AG

\lAG

0

+Fe3+

3 H f + 3L

=0

H,LLL

'I,Fe,(LL),

AG " 0

F 3H' +

LLL

+Fe3+

=o

lIAG

=O

K=1

Since the formation constant for ferric hydroxamate complexes depends only on the number of hydroxamate ligands bound to Fe and not on whether they are independent Iigands or are part of a larger multichelate molecule, equilibria such as

+

+ H2LL

2Fe(HLL)3 + Fe2(LL)3 3H2LL K =1 also have equilibrium constants near unity. In the above case, when both [FeItotaland [HzLL] are about lom3M, the ratio [Fe2(LL)3]/[Fe(HLL)3] is about lo3, since the reaction at low concentrations is strongly driven to the right. Thus for rhodotorulic acid: (1) The only tris(hydroxamate)ferric complex present in significant concentration is the dimer Fe2RA3. The lower the concentration of all species the more rigorous this conclusion becomes. Similar considerations apply to bis(hydroxamate) com-

K=1 K=1

Fe2(LL)2 + 2Fe(LL)

Both these reactions are strongly driven to the right at low concentrations and so for rhodotorulic acid: (2) The only bis(hydroxamate) ferric complex of significance is the monomer F e ( R A ) in which both hydroxamate groups are coordinated. Finally, examination of mono(hydroxamate) species and reactions such as the disproportionation 2Fe(HLL) + Fe(LL)

+ Fe3+ + H2LL K

vFe(LLL)

+ LL2-

Fe(HLL)2 F= Fe(LL)

FeL,

The small change in pK, values with successive deprotonation keeps the equilibrium constants for ligand acid disproportionation within an order of magnitude of unity, e.g. 2HLL- + H2LL

plexes, such as a 2:l complex, Fe(HLL)2, or a bis complex dimer such as Fe2(LL)2:

=

W

= 1/10

is only weakly driven to the right, nevertheless. (3) A t low concentrations of rhodotorulic acid, no stable monohydroxamate exists in significant concentration. The phenomenal production of rhodotorulic acid by R. pilimanae can now be understood. In competition with other siderophores (e.g., ferrichrome), RA is only effective at high concentration. For typical physiological concentrations of RA and ferrichrome (1.5 X lou2 and 5.5 X lom5M, respectively) and M total iron, the RA complex is favored by a factor of 2-4. Thus the high concentrations of RA produced are needed to compete effectively for iron. The biosynthesis of RA is relatively simple, and a large output of metabolic energy is not required to achieve this end.

Acknowledgment.This paper is dedicated to Professor Arthur F. Scott on the occasion of his 80th birthday. We wish to thank Professor Fred C . Anson for helpful and informative discussions and Drs. W. R. Harris and A. Avdeef for assistance with the potentiometric titrations. This research is supported by the NIH. References and Notes (1) Part 10 in this series: Carrano, C. J.; Raymond, K. N. J. Am. Cbem. SOC.

i w e , 100,5371-537a. (2) Anderegg, G.: L'Epiattenier, F.; Schwarzenbach, G. Helv. Chim. Acta 1963, 46, 1400-1409. (3) Neiiands, J. B. "Inorganic Biochemistry", G. Eichorn, Ed.: Elsevier: Amsterdam, 1972; pp 167-200.

604

Journal of the American Chemical Society

(4) Neilands, J. B. "Microbial Iron Metabolism", Academic Press: New York, N.Y., 1974. (5) Lankford, C. E. CRC Crit. Rev. Microbiol. 1973, 2, 273-331. (6) Barry, M.; Flynn, D. M.; Letsky, E. A,; Risdon, R. A. Br. Med. J. 1974, 7, 16-22. (7) Peterson, C. M.; Graziano, J. H.; Grady, R. W.; DeCiutiis, A,; Jones, R. L.; Cerami, A. 8r. J. Haematol. 1976, 33, 477-495. (8)Anderegg, G.; L'Eplattenier, F.; Schwarzenbach, G. Helv. Chim. Acta 1963, 46, 1409-1422. (9) Atkin, C. L.; Neilands, J. B. Biochemistry 1968, 7, 3734-3739. (10) Rawls, S.L. Chem. Eng. News 1977, 55, 24. (11) Avdeef, Sofen, S. R.; Bregante, T. L.; Raymond, K . N. J. Am. Chem. SOC. 1978, 100,5362-5371. (12) Bjerrum, J. "Metal Ammine Formation in Aqueous Solution", P. Haase and Son: Copenhagen, 1942. (13) The goodness of fit is defined as

a.;

(14) (15) (16) (17) (18) (19) (20) (21) (22) (23)

/ 101:3 / January 31, 1979

where A is the observed deviation between observed and calculated values, u is the standard deviation of the observation, nobsis the number of independent observations, and nvaris the number of variables optimized in the refinement. McBryde, W. A. E. Can. J. Chem. 1964, 42, 1917-1927. Martell, A. E.; Smith, R. W. "Critical Stability Constants", Vol. 4; Plenum Press: New York, N.Y., 1976. Nicholson, R. S.;Shain, I. Anal. Chem. 1964, 36, 706-723. Polcyn. D. S.:Shain, I. Anal. Chem. 1966, 38, 370-375. Anson, F. C. Personal Communication. Flanagan, J. B.; Margel, S.;Bard, A. J.; Anson, F. C. J. Am. Chem. SOC. 1978, 700,4248-4253. Neilands, J. B. fxperientia, Suppl. IX 1964, 22. Snow, G. A. 8iochem. J. 1969, 775, 199-205. Cooper, S. R.; McArdle, J. V.; Raymond, K. N. Proc. NaN. Acad. Sci. U S A . 1978, 75, 3551-3554. Anderegg, G.; L'Eplattenier, F.; Schwarzenbach, G. Helv. Chim. Acta 1963, 46, 1390-1400. For a recent discussion of the chelate effect and references to related papers, see Munro, D. Chem. 51. 1977, 73, 100-105.

Preparation and Characterization of Organoiron Secondary Alkoxycarbene Complexes Alan R. Cutler Contributionfrom the Department of Chemistry, Wesleyan Uniuersity, Middletown, Connecticut 06457. Received August 13, 1978

Abstract: A series of organoiron secondary alkoxycarbene salts, d p F e ( C 0 ) L (C(OR)HJ+PF6- (L = CO, Ph3P; R = Me, Et) (2a-d) has been synthesized by hydride abstraction from the requisite alkoxymethyl complexes (la-d) with a trityl salt. The dealkylation of 2 with iodide was investigated as a synthetic approach to ol-formyl complexes, CpFe(CO)L(CHO) (3). One equivalent of iodide dealkylated 2a-c to mixtures of CpFe(CO)zLf (6a,b), la-c, and alkyl iodide, by a reaction sequence involving hydride transfer from a transient 3 to 2. The hydride acceptor properties of 2 were also investigated by their reactions with borohydride. The observed reduction of coordinated alkoxycarbenes to alkoxymethyl and methyl ligands is of possible relevance to the catalytic fixation of carbon monoxide (e.g., Fischer-Tropsch processes) to hydrocarbons, since surface-bound secondary hydroxycarbene ligands have been postulated in forming the initial C-H bonds.

The Fischer-Tropsch synthesis and related processes' for fixing CO/H2 mixtures to methane, methanol, and higher homologues with transition metal containing heterogeneous catalysts are regaining importance as a major area of catalysis research.2 Impetus for this research activity derives from the possible use of coal, a source of these CO/H2 mixtures of "synthesis gas", as a future source of petrochemicals. Although little is known about the mechanism of CO fixation, plausible intermediates in forming initial C-H bonds often are represented as surface-bound secondary hydroxycarbene [M= CH(OH)] or formyl [M-C(O)H] ligands. Development of homogeneous analogues to this synthesis and ascertaining the role of formyl and secondary hydroxycarbene complexes in the fixation of CO correspond to recent research directions. Homogeneous hydrogenation of coordinated CO with mononuclear organometallic compounds of early transition metals may involve transitory formyl interm e d i a t e ~ . Carbonyl ~,~ ligands are also hydrogenated stoichiometrically with borohydride reagents. Thus, cationic carbonyl compounds reduce to neutral methyl5a and hydroxymethyPb complexes, and neutral carbonyl compounds reduce to anionic q1-formy16complexes. Neutral ?I1-formylcomplexes may be intermediates in the former borohydride reactions, but such compounds7 have not been detected as a result of intermolecular hydride transfer (or intramolecular hydride ligand transfer)s to a coordinated carbonyl. Secondary hydroxy- or alkoxycarbene complexes have not been reported, although many tertiary examples exist.g Information on the reactions of coordinated secondary hydroxy- and alkoxycarbene ligands clearly would be useful in probing the reaction paths available 0002-7863/79/1501-0604$01 .OO/O

during CO fixation with homogeneous or heterogeneous catalysts. We now report preparation of secondary organoiron alkoxycarbene complexes and present preliminary observations of their reactivity that may be relevant to CO fixation. Previous workers demonstrated that alkoxide abstraction from the methoxymethyl iron complexes, CpFe(CO)L(CH20CH3) (Cp = q5-C5H5),generated transient methylene iron compounds, CpFe(CO)L(CH2)+ (L = CO, Ph3P).10,11We find that Ph&+PF6- abstracts hydride quantitatively from alkoxymethyl iron complexes (la-d) in methylene chloride to give the alkoxycarbene salts (2a-d).12,'3 CpFe(CO)L-CH,

\OR -+

+

Ph3C+PF,

+,'OR .----;/

CpFe(C0)L-C

PF6- iPh,CH

H' 2 a, L = c, L =

C O ; R= Me b, R Ph,P; R = Me d, R

= =

Et Et

All four products were isolated at room temperature in 80-90% yields as stable, yellowish solids, although 2a,b were hydrolyzed slowly in air. Spectroscopic data, particularly the diagnostic 'H NMR carbene proton absorptions at low downfield position^,^^,^ and analytical evidence agree with

0 1979 American Chemical Society